1. Field of the Invention
The present disclosure relates to an information providing method, information providing device, and information providing system for accurately extracting information that is likely to attract an interest of a user.
2. Description of the Related Art
In recent years, attention has been given to a technique for collecting a user's action history and for providing information that is likely to attract an interest of the user based on the action history. The information to be collected as the action history of the user includes, for example, information regarding products purchased by the user, food ordered by the user, articles referred to by the user, and broadcast programs viewed by the user. The action history of the user may include evaluation values that each user sets on these pieces of information. The following steps show an exemplary method of extracting information that is likely to attract an interest of the user. The first step is to extract another user who has a hobby and preference with a tendency identical to a tendency of a targeted user based on the action history of the targeted user. The targeted user is a user who receives provided information. The next step is to provide the targeted user with information about a product or broadcast program that the extracted another user highly evaluates and that the targeted user has not used yet.
The following is an exemplary method of determining whether two arbitrary chosen users have an identical-tendency hobby and preference. It is determined that two users have an identical-tendency hobby and preference when the two users have similar action histories, such as ordering an identical food and evaluating the ordered food similarly.
The following is a method of accurately extracting information that is likely to attract an interest of the user. When extracting another user who has a hobby and preference with a tendency identical to a tendency of the targeted user, a user who has the identical tendency is accurately determined by calculating similarity between the users based on a characteristic vector of each user (for example, see Patent Literature 1).